Abstract More than 10,000 people per year in the United States are diagnosed with Glioblastoma multiforme, the most common primary brain tumor, and the median survival (irrespective of therapy) is one year. The multifactorial nature of this disease and limited scope of previous studies has resulted in an incomplete understanding of the etiology of glioblastoma survival. The main goal of this application is to augment the knowledge on the prognostic markers and determinants of glioblastoma survival. The conceptual framework encompasses the simultaneously consideration clinical and genomic information, use of a large sample size including 397 glioblastoma and matched control samples, and a comprehensive analysis of the relationship between glioblastoma survival and the explanatory variables using frailty models. Specific Aim 1 centers on the detection of exon expression and clinical prognostic markers of glioblastoma survival meanwhile, Specific Aim 2 entails the uncovering of microRNA expression markers together with exon expression and clinical markers. Exon and microRNA-centric and gene-centric approaches to model the genomic information will be pursued. To address the high-dimensionality of the proposed study, a combination of feature extraction strategies will be implemented. A mixed-effects proportional hazards model that accommodates for heterogeneity of variances and correlation between observations will be used to model glioblastoma survival. The posterior density estimates obtained from a Bayesian Markov chain Monte Carlo approach will offer a complete characterization of the associations between the prognostic markers and survival. Calibration of the survival predictive equation will use cross-validation approaches and benchmarking to known markers of glioblastoma survival. This application addresses several high-priority areas in cancer epidemiology research that have been identified by the NCI and our group has prior experience on the foundations of the proposed study. The outcomes of this study will be 1) the identification of clinical and genomic prognostic markers of glioblastoma survival acting independently, synergistically or antagonistically, 2) the characterization of changes in survival associated with these markers, 3) development of a more universal prediction or prognostic index of glioblastoma survival, 4) identification of glioblastoma subclasses corresponding to particular combinations of clinical and genomic levels, and 5) understanding of the interplay between a wide range of clinical and genomic factors with the ultimate goal of elucidating the etiology of cancer. The results from the proposed multi-factorial analysis of glioblastoma genomic and clinical data will support the effective design of functional molecular studies of cancer survival, facilitate the exchange of discoveries between the basic sciences and clinical and public health practice and lead to biomedical strategies to diagnose, prevent, treat and cure cancer. PUBLIC HEALTH RELEVANCE: Project Narrative This project will uncover clinical and genomic prognostic markers and determinants of glioblastoma survival. The outcomes of this study include the enhanced understanding of the interplay between multiple clinical and genomic variables with the ultimate goals of elucidating the etiology of cancer and development of a reliable survival prediction function. These findings will facilitate functional molecular studies, the exchange of discoveries between the basic sciences and clinical and public health practice and the development of biomedical strategies to diagnose, prevent, treat and cure cancer.